(navigation image)
Home American Libraries | Canadian Libraries | Universal Library | Community Texts | Project Gutenberg | Children's Library | Biodiversity Heritage Library | Additional Collections
Search: Advanced Search
Anonymous User (login or join us)
Upload
See other formats

Full text of "VIC-20 Speech Synthesiser (1982)(Adman)"

OTHER ACCESSORIES AVAILABLE 

16K Ram PacIc 
8K RamPacIc 

7P0RTExpANsi0N 
MoThERboARd 



ADMAN ELECTRONICS LTD. 
Ripon Way, Ripon Road, Harrogate, 
North Yorkshire, HG1 2AU 
Tet 0423 62642 Telex 57953 Adam G 

A MEMBER OF THE ADAM LEISURE GROUP LTD 



VIC 20 



Speech 
SyiMThesisER 

OWNERS MANUAL 



ELECTRONICS LTD 



CONTENTS 



1 ) Turn to page 5 for your Speech Synthesiser connection 
details and directions of use. 



2) Turn to page 2 for an interesting account of the speech 
synthesis technique and a fuller understanding of how 
and where to use allophones to create effective speech. 

3) Turn to page 14 for hints on how to further improve 
your articulation. 



4) Turn to page 17 if you wish to try an alternative method 
of generating speech. 



Enjoy using a fun product and by all means let us know if you 
develop any innovative applications in the home environment. 



BACKGROUND TQ ALLOPHQM6S 

Previous alternative synthesis techniques have involved synthesizing 
and storing entire words as units and unless you want to use 
excessive memory, you are limited to a small vocabulary 
For example, PULSE CODE MODULATION (PCM) which is no 
more than digital recording, storage, and playback of speech wave- 
forms, requires about 70,000 data bits/sec. of speech. Another 
method, LINEAR PREDICTIVE CODING (LPC) which predicts 
a speech sample from a weighted combination of previous samples 
requires only 1000 - 2000 bits/sec. of speech. Using this method' 
S5? ^; ox ' mate| V V 5 - 20 wo ' ds can be stored in 16K bits of memory, 
wnile these methods require prodigious memory their big advantaqe 
is relatively high quality speech. 

Allophone synthesis, on the other hand, has the major advantage 
or providing an unlimited vocabulary, since the stored units are not 
words but individual speech sounds (allophones). Each allophone 
requires a 6 bit address. Assuming that speech contains ten to 
twelve allophones/sec. allophone synthesis would require 
addressing less than 100 bits/sec. The user merely has to become 
familiar with the speech sounds of English (which are different from 
letters) and the allophone symbols used to represent them 
Another use for allophone synthesis is in a text-to-speech system. 
One limitation of allophone synthesis however, is that, although 
completely understandable, the speech quality is not as good as it is 
tor PLM or LPC, in other words it lacks intonation and character. 
Language 

To use the set of allophone sounds successfully there are a number 
of points to note. 

First there is no one-to-one correspondence between written letters 
and the sounds of a language. Each sound in a language may be 
represented by more than one letter, and conversely, each letter may 
represent more than one sound (Examples in Table 3). 
Because of these spelling irregularities one must remember to think 
in terms of sounds not letters when dealing with speech. 

The second point to be made concerns segmentation of the speech 
signal. Speech sounds are not discrete units as beads on a string are 
and cannot be called by the name of a letter. In fact speech is a 
continuously varying signal which cannot be easily broken into 
distinct sound-size units. For example, if one attempts to extract 
the b sound from the word bat by taking successively larger chunks 
of the acoustic signal from the beginning of the word, one at first 
hears a non-speech noise, and then at some point hears ba. 
In other words there is no point at which the b sound can be heard 
in isolation; one hears either a non speech noise or the syllable ba. 
Finally, the most important point to make for users of an allophone 
set, is that the acoustic signal of a speech sound may differ 
depending on whether it occurs in word-initial or word-final position 
or in the environment of a vowel which is articulated in the front or 



back of the mouth, a long or short vowel, or a voiced or voiceless 
consonant. For example the initial p in pop will be acoustically 
different from the p in spy, and may be different from the final 
p in pop. Furthermore, the ear will perceive the same acoustic 
signal differently depending on what sounds precede or follow it 
The word cot can be made to sound like cod by lengthening the ' 
duration of the o and conversely the word cod can be made to 
sound like cot by shortening the duration of the o. 
Phonemes of English 

The sounds of a language are called phonemes and each language has 
a set which is slightly different from that of other languages ' 
(Table 1 indicates characteristics of English) 

Consonants are produced by creating a constriction in the vocal tract 
which produces an aperiodic sound source. If the vocal chords are 

yi?uiVi n9 i^ , t . he ,4? me tIme as in the case of the voiced fricatives (V) 
(DHH), <DH), (2), and (ZH) (see Table 2) th ere are two sound 
sources: one which is aperiodic and one which is periodic, 
Vowels are produced with a relatively open vocal tract and a 
periodic sound source (unless they are whispered) provided by the 
vibrating vocal chords. Vowels are classified according to whether 
they are long or short, whether the front or back of the tongue is 
high or low and whether the lips are rounded or unrounded. 
In English all rounded vowels are produced in or near the back of 
the mouth, ie. OU,OUU, UH, OO, OR, OW, 

It is useful to remember that sounds which have features in common 
?m , itVItt^m^, For example the voiceless stop consonants 
(P), (T), (TT), (C), (CO, (K), require 50-80 msec of silence before 
therri and the voiced stop consonants IB), (BB), (D), (DD), and (G) 
(GG), (GGG), require 10-30 msec of silence before them 
When you find a particular technique that works well with one sound 
try using that same technique with similar sounds eg (C) sounds 

(fi° d (XRr(AV r (o n or(YR} (EE) so jt mm be used before ( "' <AA) ' 

Al lop hones 

Truthfully speaking a phoneme is an abstraction; a name given to a 
group of similar sounds in a language. Remember that (P) will be 
acoustically different depending on whether it occurs in word-initial 
or word-final position or after S. Each of these different P*s are 
allophones of the phoneme P. An allophone therefore, is what 
occurs in the actual acoustic speech signal. A phoneme is the name 
of a group of allophones. It is for this reason that our inventory 
of English speech sounds is called an allophone set. 
Extended Use of Allophone Synthesis 

The allophone set (pp 10- 1 1) contains two or three versions of some 
phonemes, ie. (H) and (HH). It is likely that you will need to use 
one allophone of a particular phoneme for word or syllable-initial- 
position. A detailed set of guidelines for using the allophones is 
given in Table 2. (suggestions not rules). 



For example, (DD) sounds good in initial position and (D) sounds 
good in final position, as in "daughter" and "collide" respectively. 
(Reference Table 4 for sample words}. One of the differences 
between the initial and final versions of a consonant is that an initial 
version may be longer than the final version. Therefore to create an 
initial S you can use two S's instead of the usual single S at the end 
of a word or syllable, as in "sister" of "sense". Note that this can 
be done with TH and F and the inherently short vowels (see below) 
but with no other consonants. You will want to experiment with 
some consonant clusters (strings of consonants such as STR 
(STRong), PL (PLuto) to discover which version of the phoneme 
works best in the cluster. (C) sounds good before (L) as in "clown", 
and (K) sounds good before (W) as in "square". One allophone of a' 
particular phoneme may sound better before or after back vowels 
and another before or after front vowels; (CO sounds flood before 
(UH) and (C) sounds good before (EE), as in "cookie". Some sounds 
(P), (B), <BB), (T), (D), (DD), (C), (CC), (K), (G), (GG), (GGG), 
(CH), and UH) require a brief duration of silence before them. 
For most of these, the silence has already been added but you may 
decide you warn to add more. Therefore there are several pauses 
included in the allophone set varying from 10 - 200 msec. 
To create the final sounds in the words "letter" and "Jittle" use the 
allophones (ER) (ERR) and (LL). The (NG) allophone obviously 
belongs at the ends of the words "sing" and "long", but notice that 
the NG sound is represented by the letter N in "uncle"; and 
remember that some sounds may not even be represented in words 
by any letters, as the |Y, in "computer", 

As mentioned earlier there are some vowels which can be doubled to 
make longer versions for stressed syllables. These are the inherently 
short vowels (A), IE), (I), (0), (U), and (UH). For example, in the 
word "extent" use one (E) in the first syllable, which is unstressed 
and two (El's in the second syllable which is stressed. Of the 
inherently long vowels there is one, (OU) which has a long and short 
version. The short one, (OU), sounds good after (Y) in computer; 
the long version (OUU) sounds better in monosyllabic words like ' 
"two". Included in the vowel set is a group called R - coloured 
vowels.^ These are vowel + R combinations clearly seen as (AR) in 
"alarm" and (OR) in "score". Of the R - coloured vowels there 
rs one (ER) which has a long and short version, the short version is 
ideal for polysyllabic words with final ER sounds like "letter", and 
the long version (ERR), is good for monosyallabic words like "fir". 
One final suggestion is the usefulness of adding a pause of 30 - 50 
msec between words when creating sentences, and a pause of 
100 — 200 msec between clauses. 



• 



CONNECTION INSTRUCTIONS 

The Speech Synthesis cartridge plugs directly into the back of 
the VIC 20 computer or into one of the slots on the 
Expansion Motherboard. 

1) Turn off the power to the VIC. 

2) Push the cartridge firmly into the expansion slot at 
the back right corner of the VIC making sure that the 
label is facing up, or in the case of our Motherboard, 
towards you. 

3) The cartridge also has 2 connection leads with integral 
5 pin connectors. The plug connector fits into the VIC 
in place of the TV modulator plug lead which itself 
connects with the second cartridge lead having a 5 pin 
socket. 

4) Turn on the power. The VIC will spring to life with the 
familiar Commodore message followed by a further 
definitive heading. 

5) The keyboard will now talk when keys are pressed. 



DIRECTIONS FOR FUN 

The operating software is initialised on power up {the operating 
system resides at SA000 - A7FF. Immediately you will find it is 
possible to initiate a response from the VIC 20 by pressing 
any of the keys on the keyboard. (Press any key now and make 
sure the TV volume is correctly set). If the F1 key from the 
right hand set of 4 function keys is pressed the letters will be 
voiced phonetically as small children learning to read would 
pronounce them. Pressing the F1 key once more restores normal 
pronunciation. 

Moving on to the creation of words from an infinite English 
vocabulary and a more limited foreign language capability we 
begin by pressing F3 key followed by the return key. Tha cursor 
should be at the left hand edge and we are now in a position to 
voice the word of our choice by entering a string of allophones. 
There are 64 allophones to choose from (see pages 10/11 ) 
For example type in the allophones H/E/LL/OO/, notice how the 
allophones develop allowing constructive alteration to take place 
at any point during the input. NB. (Each allophone must be 
followed by the separator sign /.) Alterations to a word are 
easy. Now press RETURN and disregard the syntax error 
notation. Typing in the allophones to represent the word 
CHEEKY might be attempted as CH/EE/K/EE/ (Remember that 
you must think of a word as it sounds and not as it is written). 
Suppose you wish to try the allophones C/ or CC/ in place of K/; 
use the 'delete' key 5 times and re-entaryour alternative choice ' 
of allophones, it will probably be CH/EE/CC/EE/. To separate 
words you should use a pause, the allophones P1 - P5 provide 
5 pauses of different lengths. For long intervals of silence use 



one or more of tha P5 allophone separated by "/". The pause 
allophones might be useful for deciding the batter of two 
allophone strings e.g. O/V/P5/P5/P5/0/F/. Carrying on with 
another word to follow CHEEKY you could have the following 
string CH/EE/CC/EE/P5/CH/A/TT7ER/B/0/K/S/ (you may 
return to the previous mode at any time by typing "Fl" again). 
If you continue to enter allophones a point (approx. 30 of 
tham) is reached where the allophones are no longer voiced, 
due to a lack of space in each operating storage block. 
You must now turn to the use of string formatting and the 
command SYS 41000. SYS 41000 is a key command and 
enables a string to be voiced. Begin by pressing the R UNSTOP 
and RESTORE keys simultaneously to disable the voicing of 
the keyboard when keys are pressed. The simplest program 
will illustrate the use of strings and the SYS command. 

e.g. 10 A $ - "H/E/LL/OO/" 
20 SYS 41000 

Type RUN and the word hello will be voiced. 

A further programming example might be to add the following 
lines. 



5 FOR x 
25 NEXT 



1T0 4 



Type RUN 3nd you initiate a loop sequence. 

Remember you are allowed approximately 30 allophones/ 
String. Therefore for longer sentences it is necessary to use 
more than one string, each string must have the SYS 41000 
command associated with it. 

Here is another program for you to try. First press F3 (just 
to make the entry of this program silent} 

10 FORS=1T06 

20 FORX = 1TO1200 : NEXT 

30 -AS- "CH/EE/CC/EE/" 

40 SYS 41000 

50 IFA=0THEN7O 

60 FOR D= 1 TO 1000: NEXT 

70 B S = "CH/A/TT/ER/B/O/K/S/" 

80 SYS 41000 

90 IF A = 0THEN A- 1 : NEXT 

95 IF A- 1 THEN A = 0: NEXT 

Now press RUN and RETURN for a demonstration. Repeat as 
often as you wish. You are now in a position to write your own 
voice software but may first like to look at the programs on pages 
9 & 16. 

HELPFUL HINTS. Whenever a program is RUN, the previously 
typed RUNSTOP RESTORE command is automatically lost and 
the Speech Synthesiser software is re-initialised. This can be a 
nuisance if you were originally in F1 mode because when a program 

6 



is RUN, any alteration in the program involving pressing keys 
results in the key being voiced. The answer in this case is to enter 
F3 mode before typing RUNSTOP RESTORE or after the pro- 
gram has RUN. In effect the disabling action of RUNSTOP 
RESTORE is only operable until a program is RUN. Thereafter 
the keyboard returns to the state of F1 or F3 mode. So before 
writing a program it simplifies your activities if you'enter trie F3 
mode before typing RUNSTOP RESTORE. 



A useful program for assigning allophone strings to string variables; 
typical application in a Spell and Speak program on which a 
teacher may wish to test her young students. 



(e.g. HELLOf 
(e.g. H/E/LL/OO/) 



5 FOR A = 1 TO 4 

10 PRINT "TYPE YOUR WORD" 

20 INPUT AS(A) 

30 PRINT "PRONUNCIATION" 

40 INPUT B${A) 

50 NEXT 

100 FOR B = 1 TO 1000 ; NEXT 
200 FOR C - 1 TO 4 
210 PRINT A8(C)" "; 
220 B3(C)=BS(C) 
230 SYS 41000 
240 NEXT 



The command SYS41000 will voice each string stored in turn, 

NOTE The last variable used in the program is the string that is voiced 
hence the dummy argument in line 220. 

To escape the functions provided by "F I and F3" keys press "stop/ 
restore" together (alternatively type SYS41003J (the SYS41000 
command will still work). 

Tore-initialise Fl - type SYS4 1006 
F3- type SYS41009 

To cancel FI/F3 - type SYS 41003 

(or stop/ restore) 

To voice "F1" phonetically without typing "Fl" (after initialising) 

POKE 674,128 
To voice "Fl'' normally without typing "Fl" 

POKE 674,0 

NOTE. Each string will hold approximately 30 allophones 
before running out of space. 



o 
5 



> 



I 
I 
I 



O 
O 

o 



O 



* 
* 



a 

z 

LU 
LL 

o 

1X1 

5 

LU 

Z 
O 

I 
a. 

h- 
Z 
< 

Z 

o 

(/> 

z 
o 
u 

I 

UJ 



re 
re 

£ 



x 

09 



N 



I 
O 



> 

> 



g 
< 






Q 

Q 

Q 



W N 



Z 

z 

2 



or 

EC 
[f 



» e 
*; o> 
£ Q 



x 
h 



x 
x 
Q 
l" 
Q 






O re 

.— +j 

m c 

re oi 

-I Q 



m 

2 



re 

Jo 



m T 

» I 

CD S 



5 5 



™ 15 — 

1; 13 0J 

C o o 



"S 



* -a 

u u 



tu 



— 
E 
u 



c 

> 



o 

> 



> 



o 

> 



o 

> 



Q 

> 



i 

c 



— - 






JO (M 






CD 


to — 


a 




C 







LU 


u J2 


E 




C 


*- to 
x h- 


*♦- 


1/1 




0> .- 


o 


TJ 


c 


*j c 






o 


c — 









T3 


o 






U C 

?3 


E* 


(D 


o 

Q. 


=> o 


o 




o — 

s 


> 


> 


to 


c 


*-* 


c 


* £ 


0) 

i: 


C 




S 


5 


"D 
O 
3 


E 

01 

c 




Ol 

c 


C 


o 

jz 


c 


l. 


a. 


c n 


E 


3 


01 


o n 


a 


o 


Ui 







(J 




^_ 




O 


JT 





wi 






> 


~ 


o 

c; 


a 











o 







01 


CO 


til 


"D 


a 






■01 


E 






01 


CO 




to 


jr. 


X 




— 


P 


LU 




c 







c 



x 
o 

a 

a _ 

o g 

jr ^ 

£ .9- 

Is* 

III 



(0 ,_ 

C 0! 

.E a 

x a. 

O 3 

□.■a 

a c 

|_ (D 

J3 

Irt — 

ai v) 

<J or? 

2 & 

01 S 

3 "- 



-C *j ai 



■d 5 

C ai 

to +^ 

01 01 

a a 

a a 

X) D 



c re 
•" o o 

Ol ^ 01 

3 *: > 

O Jft 

I- h- 



2S 



So 
ro o 

X 

o 

l_ 
Q. 
Q. 
CO 

<3> 

C 




5 re 
■ O i o 

.j 3 £ < 



> 
a 

CD 



re 
re 
re 
a. 



8 

o 

C9 



2 



LU 

I 
I- 



S 1 Gr 'S' 1> 

^l> "i. (Jli VjI- 

&■ S> <S:> C' 

O O O O 



r? w -* cd 

<^J Oi (M r>j 



C '^ i'J *. 






n ■* tTj i_i J 






1 1 1 1 £' S- 




cs 


xxx>:6is 




o 


Ml II II O 5) 




S 


>: x >; x — *« 






X z ^ X T t)- 




fl- ii" 


UJ UJ UJ UJ f.' ■ «,' ■ 




W Tf 


X X X X >■ > 




?- El 


1— 1— h- 1— U"> tft 




0"i — ■ 












cp : UJ (S 


/* S .- - >: \ 




Cfl N I * 


E' E> E' E> ^- ti"' 


I 


C3 * 1— 'SJ 


IP ■* iTi 01 » Q_ 


.1 


^i Q_ E' ^ 


1 1 1 1 1— - 


1 


-<- = li Q 




v- 


to H X *— 


U.1LU iLtttt 


lil 


>- W u. o 


^, M _, ^, (T (T 


Li 


trt CC •— L3 


s> o <Sf s s r> G> 


q r> £■ E> 


C« ^ <M f.* •* * 


t^ 




S) S" SJ © G» S- 


a 


,yi S> & CD 



S 



R 

S 



a 1 



\ Itl O O E* 

u5 \ © (S ® ^ A :, 

UJ *•• _-- ^t -^ > ^ -■ 

UJ _J ■*■ "* •«■ -** f* 1 ^ 

rt' i i ^^ ^- >- w w w 

111 \ (i> tfi w « ^ *-< 

NO l~ I- I- 

-*• ^ « » « « " " 
3333orir(tiEtf' 

COt?. (O-KD - CM m * "I 
plNOflf**"*' 



S 3) 

(Zl ii 

N \ 

cc to 



ii n 



» » w 3: 

lEWlh 

dt^i- 

> > > X 

I li I' ll 



It It ** *t 



6 

2* 

5- — * 

u"l CCl 
0J 
- (.n 
r- o 
v. u 
■ 

CMCS30 

— li II II 

II S> • X 

i «<r r 

ZU17IU 
UlZiUl I 
0t tf) X I- 
l-Ol-E 
CS> L3 d — • 

ii n y 
i i s: e 

u_ > U- Li. 

i-t >: •-* ~- 

ep c> fi> ca 

© — ' M t" 

ir> ir »t m 



i£> .2 a 

^- ($ Gf 

w — — 

S3 *t 
OB u"> W 
C3 » > > 
hI0 00 
CO X 

3 UJ = - 

CO X N V 

oi-ri «.■ 

= 3 



U J — * 



e a:- u ii ii to 

H U. » » » " 

>: " i i <c >- 



s> = t- 
es> \ >: 

scnui 

t-« = X 

^r h 

to M <S> 

>- <E Q 

tfj X S> 

■ LU to 

•"■ X o 

J*1H»- ffi 

w — to & 

« '*■ H * 
3 -s. OL O 
II If' Ql I- 

or — il u 



cssomciiio s b & s & 
^- r.' ifi t£ 'X- r- m q. cs -^ 
tfi IO IT) v> »"> in » ' in »o >o 



N = 
X N 

X LU 



N U. V X X X \ 









Ui t X \ N N - % 

SJ = N \ « X > J t 

= i >\S= ZLLIUJLUN 


















- 








01N= „\\I-H|I1J31I 








5i\HMwfwJ:\J) 








oua\\ ^^ N,_u i! _ rE 


- 






T s o (i. l/i if 1 1 >- ""- >- 








E ? ul u. t/i to F = JL A X X 

"ll "ll ~ll "ll "* "ll ^ — S) -- CM CT> 


CO 


o 

II 


A 




^"\ 


,-^ ,^_ r^ ^*\ ^*^ ^^ CO Cfl »-4 *-t •-< »-■ 



J LU Id LU I- 
n. x LU UJ \ 



\ * i3\ \« "\ 

X N \ t- U V LU 
\ XH I- LU LULU 
LU S fr-N. N &t N 

\ UJ \ x i— * L7 f- 

"^ LU 7 x - I" 1" H 
\ \. S LU \ I- N 

iijl-«NLV Mt 
\ HM3LUCL \ 
if, X N V \ O " 

x tr x i- x x x 

m i r i- t- u. u. 

ii ii 



i r_ oo cfi ts *- cm w 
, Ui " — (M r»j w r»j 



* " JL * 



x>-v-:isr>i ^iiiriL_^i-i-i-Ht-hi-i-i-i-i-i-i- 



^ i 
CB 



■'■'"'"'"'"'"(SCSIBBS.tSOpOOOOgSigS 



= 


c 


in 


s 


s 


S 


n 


»~i 


IO 


UJ 


H-t 


X 




s 




J . 


1- 


v. 


a 


1- 


" 




II 


II 








-4 


fj 


r. 








w 


<n 


B 


-! 


01 


J 


1 


s s> 


'J 


r^ 


n 


1-1 


w 


s 



8 



9 



fSPEECH SYNTHESISER ALLOPHQNE TABLE~j 



Allophone 


Sample Word 


Duration 




PI 




10 MS 




P2 




30 MS 




P3 




50 MS 




P4 




100 MS 




P5 




200 MS 


I 


• A 


HaT 


100 MS 


AA 


BeiGE 


2S0 MS 




*E 


eND 


70 MS 


) 


EE 


See 


250 MS 


• I 


SiT 


70 MS 




II 


SKy 


260 MS 




•o 


HoT 


100 MS 




00 


Beau 


240 MS 




*u 


Succeed 


70 MS 




B 


bOTTLE 


50 MS 




BB 


bUSY 


80 MS 




C 


CAN'T 


120 MS 




CC 


cOMB 


190 MS 




D 


COULd 


70 MS 




DD 


dO 


160 MS 




*F 


phONE 


150 MS 




G 


gUEST 


40 MS 




GG 


gOT 


80 MS 




GGG 


Wig 


140 MS 




H 


hE 


130 MS 




HH 


hOE 


180 MS 


) 


J 


DOdgE 


140 MS 


K 


SkY 


190 MS 


5 


L 


1AKE 


110 MS 


LL 


SADDle 


190 MS 




M 


mILK 


180 MS 




N 


THIn 


140 MS 





Allophone 


Sample Word 


Duration 


NN 


nO 


190 MS 


P 


pOW 


210 MS 


R 


BrAIN 


120 MS 


RR 


rURAL 


170 MS 


*s 


BEst 


90 MS 


T 


PARt 


100 MS 


TT 


to 


140 MS 


V 


SEvEN 


190 MS 


W 


wET 


180 MS 


Y 


yES 


130 MS 


YY 


yEAH 


180 MS 


Z 


7.00 


210 MS 


DH 


rauM 


240 MS 


DHH 


merM 


290 MS 


OR 


STore 


330 MS 


AR 


ALarM 


290 MS 


YR 


CLear 


350 MS 


OY 


Boy 


420 MS 


OU 


LuKE 


100 MS 


OUU 


FooD 


260 MS 


NG 


AnGER 


220 MS 


OW 


ouT 


370 MS 


*EH 


WarY 


120 MS 


ER 


Fir 


160 MS 


ERR 


Purr 


300 MS 


*TH 


thIN 


180 MS 


WH 


whIG 


200 MS 


CH 


chURCH 


190 MS 


SH 


shIP 


160 MS 


*UH 


TooK 


100 MS 


XR 


REPair 


360 MS 


ZH 


AzURE 


190 MS 



10 



-%■ These allophones may be doubled 
Remember to separate with a / sign 



11 



TABLE 2 - GUIDELINES FOR USING THE ALLOPHONES 

•• VOICED STOP CONSONANTS 

/B/ - final position: rib; between vowels: 

fibber; in clusters: baeed, brown 
/BB/ - initial position before a vowel: beast 
/D/ - final position: played, end 
/DD/ - initial position: down; clusters: drain 
/G/ - before high front vowels: YR, IY, IH, 

EY, EH, XR 
/GG/ - before high back vowels: UW, UH, OW, OY, 

AX; and clusters: green, glue 
/GGG/- before low vowels: AE, AW, AY, AR, AA, AO 

OR, ER; and medial clusters: anger; and 

final position: peg 

•• VOICELESS STOP CONSONANTS 

/P/ - pleasure, amp_le, tri£ 

/!/ - final clusters before SS: tests, its 

/TV/ - all other positions: t_est_, street^ 

/C/ - before front vowels: YR, IY,~IH,~EY, EH, 

XR, AY, AE, ER, AX; initial clusters: 

cute, clown, scr eam 
/K/ - final position: speak; final clusters: 

task 
/CC/ - before back vowels: UW, UH , OW, OY, OR, 

AR, AO; initial clusters: crane, quick, 

clown, scr eam 

•• NASAL 

/N/ - before front and centre vowels: YR , IY, 
IH, EY, EH, XR, AE, ER , AX, AW, AY, UW; 
final clusters: earn 

/NN/ - before back vowels: UH, OW, OY, OR, AR, AA 

/M/ - milic, alarm, ample 

/NG/ - string, anger 

••SILENCE 

PI (10 MS) - before BB , DD, GG, and JH 

P2 (30 MS) - before BB, DD, GG, and JH 

P3 (50 MS) - before PP , TT , KK, and CH, and 

between words 

P4 (100 MS) - between clauses and sentences 

P5 (200 MS) - between clauses and sentences 

•• AFFRICATES 

/CH/ 

/J/ 



c h u r c h , feature 
judge, injure 






•• VOICED FRICATIVES 

/V/ - vest, prove, even 

/DH/ - word-initial position: _this, then, they 

/DHH/- word-final and between vowels: bath" 

bathing 
/Z/ - zoo, phase 
/ZH/ - beige, pleasure 

•• VOICELESS FRICATIVES 

*/F/ - ) 

*/TH/- ) The5e ma Y be doubled for initial posi- 

*/S/ - ) tion and used singly in final position 

/SH/ - shirt, leash, nation 

/H/ - before front vowels: YR , IY, IH, EY EH 

XR, AE ' 

/HH/ - before back vowels: UW, UH, OW OY AO 

OR, AR ' ' 

/WH/ - white, whim, twenty 

•• RESONANTS 



/w/ - 
/R/ - 
AR/ - 
A/ - 
tV - 

/YY/ - 

•• LONG 

/EE/ - 
/AA/ - 
/II/ - 
/OY/ - 
/OU/ - 
/OUU/- 
/OO/ - 
/OW/ - 
AL/ - 



we, warrant, linguist 

initial position: read, write, x-ray 

initial clusters: b£own, C£ane, g~ease 

.like, hello, steel 

clusters: cute, bea uty, computer 

initial position: jes, ^arn, yo-yo 

VOWELS 

treat, people, penny_ 
great, statement, tray 
kite, sky_, tnighty 
noise, toy, voice 
after clusters with YY 
in monosyllabic words: 
zone, close, snow 
sound , mouse, down 
little, angle, gentlemen 



computer 
two, food 



*THESE ALLOPHONES MAY BE DOUBLED 
1? 



•• R-COLOURED VOWELS 

/ER/ - lette_£, furniture, interrupt 

/ERR/- monosyllables: bird, fern, burn 

/OR /- fortune, adorn, store "~ 

/AR/ - farm, alarm, garment 

/TO/ - hear, earring, irresponsible 

/XR/ - h air , decl are , st are 

••SHORT VOWELS 

*/I/ - sitting, stranded 
*/E/ - extent, gentlemen 
*/A/ - extract, acting 
*/UH/- cookie, full 
*/EH/- air, 
*/U/ - lapel, instruct 
*/0/ - pottery, cotton 

13 



HINTS FOR USING THE ALLOPHONES 

We have designed the allophone table in such a way that you can 
quickly and confidently expect to produce excellent speech in a 
very short time. Remembering the allophones provides you with 
the quickest method of producing results and in most cases you will 
find that there is no need to continually refer to the table as you 
progressively improve your ability to manipulate the allophones. 
There will be, however, occasional instances in which the apparently 
correct allophones do not produce the required word. It is here 
that one has to experiment, often with some highly unusual or 
improbable allophones from the master list, and to illustrate these 
conditions there are a number of alternatively programmed words 
below. Try them and see the difference! Place them on the same 
line and separate each alternative with a pause. 



P/R/EE/S/I/SH/U/N/ 

P/ER/E/S/l/SH/U/rg/ 

B/ER/G/A/N/D/EE/ 

BB/EER/GGG/ER/ISI/DD/EE/ 

L/E/NG/TH/ 

LL/E/N/TH/ 

LL/E/N/C/TH/ 

V/E/ER/I/F/II/ 

F/E/ER/I/F/II/ 

V/ER/I/F/A/II/ 

H/ER/CH/ 

CH/ERR/CH7 

CH/ER/CH/ 



- looks correct ) 



JPRECISION 



-sounds correct ) 



—looks correct ) 



IBURGU'NDY 
—sounds correct ) 

—sounds poor ) 

) 
-sounds correct ) LENGTH 

) 
—sounds correct ) 

—sounds poor ) 

I 
—sounds poor )VERIFY 

) 
—sounds correct ) 

—sounds poor ) 

) 
—sounds similar JCHURCH 

) 
—sounds correct ) 



TABLE 3 - SPELLING IRREGULARITIES 





One Sound 
to Many Letter 
Representation 


Many Sound 
to One Letter 
Representation 


VOWELS 


mejt 

fr^et 

Pete 

people 

penny_ 


vein 

foreign 

deism 

deicer 

geisha 


CONSONANTS 




ship 
tension 

precjous 
nation 


although 

ghastly 

cough 



TABLE 4 - EXAMPLES OF WORDS MADE FROM ALLOPHONES 



IM.B. (FUNC 3) Should you find lhai when you input allophones and 
complete a line there is no answering sound do not be alarmed; this 
is an operating characteristic. Delete once and retype "/" if vou wish. 
Sound effects may be created by concatenating a series of allophones 
of varying length. 



e.g.D/D/D/D/D/ 



G/G/G/G/G/ 
K/K/K/K/K/ 



Z/Z/Z/Z/Z/ M/M/M/M/ 



DD/P1/0/0/TT/ER/ 

C/0/L/II/P1/D/ 

S/S/I/S/TT/ER/ 

C/L/OW/N/ 

CC/UH/C/EE/ 

L/E/TT/ER/ 

L/I/TT/LL/ 

U/NG/CC/LL/ 

C/O/M/P/Y/OU/TT/ER/ 

E/CC/S/TT/E/E/N/T/ 

TT/OUU/ 

A/LL/AR/P1/M/ 

S/CC/OR/ 

F/ERR/ 



"daughter" 

"collide" 

"sister" 

"clown" 

"cookie" 

"letter" 

"little" 

"uncle" 

"cnmputRr' 

"extent" 

"two" 

"alarm" 

"score" 

"fir" 



14 



15 



HEAR ALL THE ALLOPHONES 



1 PRINT"" 

2 PRINT" I WILL LET YOU HEAR EACH RLLOPHOHE 
5 FQRA-0TO2000 NEXT 

4 V=40959 DIMR*<;64.v0iriB*<64? 

5 cosueieee 

10 F0RA-5T063 

11 PRINT" "^ PRINT PRINT PRINT" ALLOPHONE # "fl 

12 PRINT PRINT PRINT" NAME "A*(A> 

13 PRINT PRINT <PR I NT" SAMPLE WORD "B$CR> 

18 FORK=9TO2'URITV,l-POKEV,R'WflITV,l<POKEV.0 

31 IFPEEK<203)><64 THEN31 

33 FORL*0TC600- NEXT: NEXT NEXT 

90 FORJ=0TO7000 NEXT 

180 FORH=0TO23READI WAITV, I POKEV, I NEXT 

399 END 

1000 F0RF=5T0S3REfiDAi<.F> NEXT 

1039 F0RG=3T063 RERDBK G ) ■ NEXT > RETURN 

2000 DATA/OY/,/II/,/E/,/CC/,/P/,/J/,/N/,/I/,/TT/ 

2010 DRTA/R/,/U/, /M/, /T/> /DH/ ,/EE/, /AR/, /DA /OUV 

2015 DAT R/O/, /fl/ 

2020 DATfl/YY/ , /gH/ , /H/ , '8/, /TH/, /UH/, /OUU/ . /OU/ 

2023 data/dd/,.-ggg/ 

2930 drta/v/ , .'gg/ , /sh/ , /zh/ , /rr/ , /f/ , /*/ , /c/ , /z/ 

2033 DATfi/NG/ . AJ 

2040 DAT A/14/, /XR/, /WH/, /Y/, 'CH/, /ERA /ERR/ * /0OV 

2043 DRTR/DHH/, /S/ , /Nr)/ , /HH/, /OR/ 

2050 DATA/AR/, /YR/, /G/, /LL/,/BB/ 

2300 DATABOY , SKY , END, COMB ,P0W, DODGE, THIN, SIT, TO 

2510 DATARURAL 

2330 DATA$UCCEED,MILK, PART, THEY- SEE, BEIGE, COULD, TO 

2555 DRTRRUGHT 

2560 DATRHAT, YES, WARY, HE, BUSINESS, THIN, BOOK, FOOD 

2370 DRTflOUT,DO,UIG,VEST 

3000 DATAGOT , SHIP, AZURE, BRAIN, FOOD, SKY, CRN' T 

30 1 O DRTRZOO , ANCHOR , LAKE , WOOL , REPA I R , WH I G , YES 

30 1 5 DAT ACHURCH , F I P , F 1 PR ■ OH , THE Y - VEST , NO , HOE 

3020 DAT ASTORE , RLARP1 , CLEAR , GUEST , SADDLE , BUS I NESS 

4000 DRTR27, 7, 43, 33, 4, 16, 6, 11, 20, 16, 3, 12, 38, 2, 30 

4910 DATR24, 13,51 , 2, 63, 23, 55, 35, 






Copyright S.T. Currah 1982 



ALTERNATIVE SPEECH GENERATION 
Users may be interested to try this mode of operation. The method 
ot programming speech using this mode relies on the user becoming 
acquainted with the allophone list on pages 18/19 It requires the 
user to be more patient and may take longer to construct speech 
than the previous mode does, but it is still possible to draw upon 
an infinite vocabulary when you manipulate the allophones. 
Press RUNSTOP RESTORE or type SYS410O3 to disable the 
existing software. For a return to previous mode refer to SYS 
commands on page 7. For example, a word such as "answer" 
would be voiced by typing in 26,1 1,55,52. However, just enter- 
ing those numbers as they stand will not initiate any sound The 
procedure for entering the chosen word or words is to provide 
the following program format: 

100 FOR A-0TO23 

110 READ D 

120 WAIT 40959, 1 

130 POKE 40959, D 

140 NEXT A 

150 DATA 27,7,45,53,4,16,6,11,20 16 3 12 

38,2,28,24,11,43,53,4 
HELLO 27,7,45,53 

SPACE 4 

MY NAME 16,6,0,11 2016 
SPACE 3 

IS 12,38 

BONZO 28,24,11,43,53,4 

N.B. It is important to put a pause at the end of your chosen 
allophones to prevent the last allophone continuing. You might 
decide at this point to voice the short string several times This 
may be performed by including the following new lines in the 
preceding program. 

90 FOR B=1 TO 3 

145 RESTORE 

146 NEXT B 

Such is the programmability of the module. 

WAIT 40959, 1 This is used to ensure that the last 

allophone has been voiced before 

proceeding. 

POKE 40959, D D is the databyte for the allophone. 

Should you wish to use the words "HELLO", "MY", "NAME", 

IS , "VIC" in randomly selective manner, as you would in 
games software, you must apply the WAIT and POKE routine 
to each word or words In order to then program in the random 
selection. 



ifi 



17 



CODED ALLOPHONE TABLE 



DECIMAL 
CODE NO 



1 

2 

3 

4 
26 
20 

7 
19 
12 

6 
24 
53 
15 
ft 3 
28 

8 
42 
41 
21 
33 
40' 
61 
36 
34 
27 
57 
10 
45 
62 
16 
11 



ALLOPHONP. 


SAMPLE WORD 


DURATION 


PI 




10 MS 


P2 




30 MS 


P3 




50 MS 


P4 




100 MS 


P5 




200 MS 


*A 


IlaT 


100 MS 


AA 


BeiGE 


2 SO MS 


*E 


eND 


70 MS 


EE 


See 


250 MS 


*I 


SiT 


70 MS 


II 


SKy 


260 MS 


*o 


HoT 


100 MS 


00 


Beau 


240 MS 


*u 


succeed 


70 MS 


D 


bOTTLE 


50 MS 


BB 


bUSY 


80 MS 


c 


CAN'T 


120 MS 


CC 


cQMB 


190 MS 


D 


COULd 


70 MS 


DD 


dO 


160 MS 


*F 


phONE 


150 MS 


C 


gUEST 


40 MS 


GG 


gOT 


80 MS 


GGG 


Wig 


140 MS 


H 


hE 


130 MS 


HH 


hOE 


ISO MS 


J 


DOdgE 


140 MS 


K 


SkY 


190 MS 


L 


1AKE 


110 MS 


LL 


SADDle 


190 MS 


M 


mILK 


180 MS 


N 


THIn 


140 MS 



DECIMAL 
CODE NO 

56 

9 
39 
14 
55 
17 
13 
35 
46 
49 
25 
43 
54 
18 
:■■* 
59 
60 

5 
22 
31 
44 
32 
26 
51 
52 
29 
48 
50 
37 

30 
47 
38 



ALLOPHONI : SAMPLE WORD DURATION 



NN 


nO 


190 MS 


P 


pOW 


210 MS 


R 


BrAIN 


120 MS 


RR 


rURAL 


170 MS 


*S 


BEst 


90 MS 


T 


PARt 


100 MS 


TT 


to 


140 MS 


V 


SEvEN 


190 MS 


w 


wET 


180 MS 


Y 


yES 


130 MS 


YY 


yEAH 


180 MS 


2 


zOO 


210 MS 


DH 


muM 


240 MS 


Dim 


merM 


290 MS 


OR 


STore 


330 MS 


AR 


ALarM 


290 MS 


YR 


CLear 


350 MS 


OY 


Boy 


420 MS 


OU 


LuK.F 


100 MS 


OUU 


FooD 


260 MS 


NG 


AnGER 


220 MS 


OW 


OuT 


370 MS 


#£H 


WarY 


120 MS 


ER 


Fir 


160 MS 


ERR 


Pu r r 


300 MS 


-*TH 


thlN 


180 MS 


WH 


whTG 


200 MS 


CH 


chURCH 


190 MS 


SH 


shlP 


160 MS 


*UH 


TooK 


100 MS 


XR 


REPair 


360 MS 


ZH 


AzURE 


190 MS 



-^ These allophones may be doubted 



19 



NOTES 

1 ) Any unauthorised modff ication of these cartridges voids the 
guarantee, 

2) Only one voice synthesis cartridge may be used at any one time 
in the VIC Using a motherboard, memory RAM packs may be 
used alongside our speech units for further storage allowance. 

3) Take care not to drop the units and avoid touchina the metal 
contacts on the connector to prolong unit life. 

4) We offer a 12 month guarantee on each product and will replace 
any defective unit returned to us accompanied by the 
appropriate safes receipt. 

5} If a disc drive is to be used it is essential that the disc drive 
be turned on before the VIC is powered up. 

6) If the module is used with the VIC Programmers Aid 
Cartridge use SYS 41009 if in F3 Mode. 

7) Conflicts may arise when used with other cartridges except 
RAM packs. 



ACKNOWLEDGEMENT 

Thanks to General Instruments for applications assistance. 
Sections of this manual are "Extracted from General Instruments 
Applications Report". Copyright 1982 

20